Lunyiu Nie

Efficient Agentic AI.
PhD @UTAustin, Looking for full-time roles.
Intern @Meta MSL, ex-Adobe Research, Microsoft Research
Prev. @Tsinghua_Uni @CUHKofficial.

[Lunyiu Nie @flitternie]

Current RL post-training may not suit open-ended research. What would it take to train AI researchers to develop the research taste needed to decide what to pursue, when to pivot, and where to invest next?

Read my blog on AutoResearch: https://t.co/HcjL7g6Lur

[Lunyiu Nie @flitternie]

Jev has attracted many attention by treating decision-making as a first-class model problem.

I wrote about where it fails, what its probabilities actually mean, and why it may matter for enterprises and long-horizon agents for open-ended tasks.

https://t.co/Ek4cv6lCt0

[Lunyiu Nie @flitternie]

(4/4)
The results speak for themselves:
◆ 5x speedup at the same report quality
◆ 4.1x throughput improvement
◆ 3.39x cost reduction via early pruning

Long-horizon agents need efficiency as much as intelligence.

See you in Rio 🇧🇷 come find me at the workshop or just say hi!

[Lunyiu Nie @flitternie]

I made Deep Research 5x faster — without sacrificing quality.

Excited to share that ParallelResearch, my internship work at Adobe Research, was accepted as a 🌟 Spotlight at the ICLR 2026 Workshop on Agents in the Wild (AIWILD).

Here’s the idea 👇

[Lunyiu Nie @flitternie]

(3/4)
Here are the keys:

🌳 Adaptive planning: expand only when information gain justifies the cost
⚡️ Real-time orchestration: prune low-value branches mid-research, reallocate compute on the fly
🚀 Fully async execution: no more waiting on slow dependencies

[Lunyiu Nie @flitternie]

@jiefengchen1 Hi Jiefeng! I'm a CS PhD at UT Austin, with focus on agent workflow orchestration & code intelligence (https://t.co/o26tuauadk). I have passed the general interview and am currently on Google's team match phase. Just sent you my CV in email. Would love to talk more!

[Lunyiu Nie @flitternie]

I re-plotted the CursorBench results against current API pricing. GPT becomes the new coding king?

Calculation: (Median Tokens / 1M) * Blended Rate.

Assumes a 3:1 input/output ratio plus reasoning token overhead. Analysis via Gemini.

[Cursor @cursor_ai]

We're sharing a new method for scoring models on agentic coding tasks.

Here's how models in Cursor compare on intelligence and efficiency:

[Lunyiu Nie @flitternie]

Huge thanks to my wonderful co-authors @ZMDing99, Kevin Yu, Marco Cheung, Chris Jermaine, and @swarat!

[Lunyiu Nie @flitternie]

1/2 Excited to share our @COLM_conf paper on making Agentic AI systems more cost-efficient in real-world deployments.

We break multimodal reasoning tasks into Foundation Model Programs that learn to dynamically route subtasks to the best backends.

📄 https://t.co/eBVYhv17Ds

[Lunyiu Nie @flitternie]

2/2 Results: up to 98% cost savings with minimal loss in task performance — without locking into one monolithic model.

We also release new benchmarks for streaming VQA on resource-constrained settings. Check it out on our website.

🔗 https://t.co/1J5B26XisQ

[Fangcong Yin @fangcong_y10593]

Solving complex problems with CoT requires combining different skills.

We can do this by:
🧩Modify the CoT data format to be “composable” with other skills
🔥Train models on each skill
📌Combine those models

Lead to better 0-shot reasoning on tasks involving skill composition!

[Swarat Chaudhuri @swarat]

Partial group reunion at #ICML2024.

[Lunyiu Nie @flitternie]

🙌 Big thanks to my amazing co-authors @ZMDing99, Erdong Hu, Christopher Jermaine, and my advisor @swarat for their incredible support and contributions to this research!

[Lunyiu Nie @flitternie]

🚀 Thrilled to present our paper "Online Cascade Learning for Efficient Inference over Streams," to appear at #ICML2024! 🎉 We've crafted a new way to switch between LLMs and cheaper models learned online, significantly reducing inference costs. https://t.co/0jYBknUCyF

[Lunyiu Nie @flitternie]

🌍 We believe this work opens a door for stream applications where *cost and efficiency* are crucial. In long run, we are building a system that can optimize the coordination architectures among models/agents under cost constraints, for complex reasoning. Follow us for updates!